Newest papers

Where is Physical RSI capability from? An Observational Scaling Law of LLM-as-policy

Zekai Li, Jiayi Yu, Jinlu Zhang, Luyang Niu, Zekun Qi, Jun Zheng, Xuchuan Chen, Dairu Liu · 2026 · arXiv

DOI: 10.48550/arXiv.2610.26848 · open-access PDF

Spanish

En este artículo, investigamos sistemáticamente los mecanismos, el comportamiento de escalado y la generalización de la mejora física recursiva. Introducimos PhysRSI-eval para cuantificar el RSI físico a través de una métrica de cinco ejes: primer intento de rendimiento, eficiencia, éxito, presupuesto y mejora del fracaso. Los diferentes VLMs se evalúan en 18 tareas de manipulación bajo un presupuesto de interacción fijo. El análisis de los resultados experimentales motiva la hipótesis de barril físico: diferentes capacidades pueden restringir diferentes partes de la mejora. Las métricas de cinco ejes también distinguen sistemas con público cercano

Machine translation (Meta M2M100); the English original follows.

Abstract

In this paper, we systematically investigate the mechanisms, scaling behavior, and generalization of physical Recursive self-improvement. We introduce PhysRSI-eval to quantify physical RSI through a five-axis metric: First-Attempt Performance, Efficiency, Success, Budget and Improvement from failure. Different VLMs are evaluated on 18 manipulation tasks under a fixed interaction budget. The analysis of experimental results motivates the Physical Barrel Hypothesis: different capabilities may constrain different parts of improvement. The five-axis metrics also distinguish systems with close publ

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.26848, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange